SLAs from HP and AWS shouldn't make you feel particularly confident

By  

If Gartner's Lydia Leong is right, Amazon and HP have little confidence in the reliability of their own cloud services.

That's my take after reading Gartner analyst Lydia Leong's blog post that examines the SLAs offered by Amazon Web Services and HP, which just announced the SLA for its Cloud Compute service. Leong called it "debatable if this is even worse than AWS EC2 SLA," via Twitter.

She argues that it's clear SLAs aren’t in place to reimburse customers for losses they might incur while the systems are down. "Rather, the monetary guarantees involved are an expression of shared risk. They represent a vote of confidence — how sure is the provider of its ability to deliver to the SLA, and how much money is the provider willing to bet on that?" she wrote in a blog post.

If that's true, then apparently AWS and HP aren't very confident in their services. Otherwise, why would the write their SLAs to make customers jump through hoops to get any return after an outage?

HP, which Leong comments had an opportunity to do better than AWS but clearly decided not to, defines an outage as something impacting every availability zone in a region, which isn't likely to happen, she wrote.

Amazon essentially forces companies to replicate their data in two availability zones in a region.

Leong found that AWS and HP aren't the norm. Most IaaS providers have "decent givebacks" and don't require users to replicate in multiple data centers. She called out Dimension Data for the simplicity of its SLA, which is calculated on a per instance basis and is calculated monthly.

I was curious how Rackspace, which just launched its OpenStack cloud this summer, measured up. It bases its SLA on "cloud server hosts," which it defines as the physical server that hosts a customer's cloud server. That doesn’t sound like a per-instance arrangement.

I asked Rackspace how its SLA measured up to others and it pointed me to the SLA page and offered some boilerplate comment about its support services.

There's likely more debate on the subject to come. Via Twitter, HP told Lydia it disagreed with her take and offered more information. Leong replied that she's likely to follow up with a more detailed comparison after discussion with her colleagues.

Leong left her readers with a couple nuggets of advice: "expect that the likelihood of a meaningful giveback is basically nil" and buy cyber-risk insurance to mitigate potential huge losses.

Join us:
Facebook

Twitter

Pinterest

Tumblr

LinkedIn

Google+

Cloud ComputingWhite Papers & Webcasts

See more White Papers | Webcasts

Answers - Powered by ITworld

ITworld Answers helps you solve problems and share expertise. Ask a question or take a crack at answering the new questions below.

Join us:
Facebook

Twitter

Pinterest

Tumblr

LinkedIn

Google+

Ask a Question